KOF Working Papers

Keyness Aggregate Supply Function: everything you always wanted to know about Z

Jochen Hartwig

No. 282 May 2011

Electronic copy available at: http://ssrn.com/abstract=1855728

ETH Zurich KOF Swiss Economic Institute WEH D 4 Weinbergstrasse 35 8092 Zurich Switzerland Phone + 41 44 632 42 39 Fax + 41 44 632 12 18 www.kof.ethz.ch kof@kof.ethz.ch

Electronic copy available at: http://ssrn.com/abstract=1855728

Keyness Aggregate Supply Function: everything you always wanted to know about Z

Jochen Hartwig KOF Swiss Economic Institute at ETH Zurich Weinbergstr. 35, 8092 Zurich Switzerland

Telephone: Fax: Email:

+41-44-632-7331 +41-44-632-1218 hartwig@kof.ethz.ch

JEL classification: Keywords:

B31, C43 Keynes, aggregate supply function, output heterogeneity

Abstract The paper suggests a consistent interpretation for the much debated Z-footnote on pp. 55-56 of the General Theory and discards claims recently made in the literature concerning the importance of output heterogeneity for Keyness macroeconomic approach.

1.

Introduction

Somethings wrong with footnote 2 on pp. 55-56 of Keyness General Theory, so it seems. In that footnote, Keynes makes two seemingly contradictory statements about the slope of the aggregate supply function Z. First he says that the slope of Z in wage units (Zw) is equal to 1. In the last sentence of the footnote however, he states that the slope of the aggregate supply function is equal to the reciprocal of the money wage. In a recent debate on the footnote both Hayes (2007, 2008) and Hartwig and Brady (2008) were able to make sense of the first part of the footnote by showing (albeit in quite different ways) that the slope of Zw equals one. Also, both Hayes and Hartwig and Brady came to the conclusion that the last sentence of the footnote must be wrong. Gerhard Michael Ambrosi has now joined the debate, advertising his interpretation as a candidate for settling the long 1

dispute over what Keynes meant in the footnote (Ambrosi, 2011, p. 633). Ambrosi in principle agrees that the last sentence is wrong but claims that it could be corrected by inserting the word share at the very end of the footnote. In other words, Ambrosi thinks that Keynes really meant that the slope of the aggregate supply function is equal to the reciprocal of the money wage share. In this paper I will argue that although Ambrosis contribution indeed pushes the debate forward into the right direction, there is still room for improvement and clarification. Ambrosis claim that the last sentence of the footnote has something to do with the inverse of the wage share is indeed correct, as is his insistence that the first part of the footnote and the last sentence deal with different curves. In other respects, however, there are a number of misrepresentations and even false claims in Ambrosis article which need to be rectified. One such misrepresentation concerns his emphasis on the heterogeneity of output.

2.

Heterogeneous output: does it matter for macroeconomics?

Ambrosi joins Hayes (2007, 2008) in emphasizing the heterogeneity of output and the alleged inadmissibility of dealing with concepts such as aggregate output or the economy-wide production function.1 Hartwig and Brady (2008), on their part, while acknowledging that Keynes expresses his unease with concepts like aggregate levels of prices and output in chapter 4 of the General Theory, plead for common sense. They argue that it would be unwise to drop a useful construct like real GDP from macroeconomic analysis. Ambrosi clearly misrepresents their argument when he insinuates that they are assuming a central planning bureau instead of acknowledging that the economy consists of a multitude of profitmaximising firms. In earlier papers on Keyness principle of effective demand see Hartwig (2007) for just one example I repeatedly addressed the question how the macroeconomic model of supply and demand from chapter 3 of the General Theory can be construed taking profit maximisation of individual firms as starting point.2 This is a non-trivial problem,
1

Apparently, however, Ambrosi and Hayes draw different conclusions for the aggregation of firms

heterogeneous output. Ambrosi (2011, p. 628) arrives at the macroeconomic Z function by summing firms individual Z functions. Hayes, on his part, writes: It is natural to think of the aggregate supply function being built, from the bottom up as it were, as the sum of the aggregate supply functions of individual firms: based upon the microfoundation of industry supply and demand. Keyness method runs the other way, from top down, from the aggregate supply function of industry as a whole, to those individual industries: truly macrofoundation of microeconomics (Hayes, 2007, p. 744).
2

Ambrosi, in his footnote 8, displays a curious understanding of the term macroeconomic when he claims that

only one-sectoral models of the overall economy are macroeconomic models. This is of course not the case. Also

especially for the aggregate demand function (D), in my opinion. For the aggregate supply function, Z, however, I tend to share Ambrosis (rather than Hayess) view that the individual firms Z functions can be summed to arrive at the macroeconomic Z function. Of course one must measure the individual firms outputs in terms of money in order to be able to aggregate them. But once we have summed the firms sales and deducted their purchases of intermediate inputs to arrive at nominal GDP so to speak, we can deflate this magnitude appropriately in order to separate the price and the quantity component in GDP or, more precisely, the change in these two components from one period to the next. Then it is also possible to construct an index representing the overall price level which has the value of 100 in an arbitrarily chosen base period and an index representing the overall level of real output which has the same value as nominal GDP in the base period. This is standard practice in statistical offices around the world. Nowhere along the way do we need to assume a central planning agency or that every firm produces the same product. No-one denies that output is heterogeneous in the real world. The point is that we can construct a homogenous measure for the aggregate of heterogeneous real output with which we can do macroeconomic analysis. Froyen (1976), in his study on the aggregative structure of the General Theory, draws attention to the fact that Keyness main concern was explicitly with the Theory of Output and Employment as a whole (Keynes, 1936, p. 293). Froyen (1976, p. 381) concludes that considerations concerning the contrast between the one-good model and the n-good economy the model represented were of secondary importance for Keynes because he ignored them in the formal model of the General Theory, which consists of only four aggregates: commodities, money, labour services and nonmoney assets, and three relative prices: the level of commodity prices, the interest rate, and the money wage. As a matter of fact, Keynes uses the concepts of aggregate real output (O) and the aggregate price level (P) several times in the General Theory. Normally, he adds warning notices such as: if it is practicable to measure the quantity, O, and the price, P, of current output (Keynes, 1936, p. 209) or (i)f we are dealing with industry as a whole and are prepared to assume that we have a unit in which output as a whole can be measured (Keynes, 1936, p. 285). In the remainder of this paper, I will assume that this is indeed practicable.3 Following Chick (1983, p. 66), I
a two-sectoral model (like, for instance, Pigous model which Keynes discusses in the appendix to chapter 19 of the General Theory) is a macroeconomic model and the corresponding production functions of the wage goods and non-wage goods sectors are macroeconomic production functions as opposed to microeconomic production functions of single firms.
3

Let us not forget that at the time Keynes wrote the General Theory, National Accounting did not yet exist. As a

matter of fact, Keyness theory was an important inspiration for economists like Nobel laureates James Meade

will therefore assume that Z(N) is derived from the aggregate production function

O ( N ) , keeping in mind that this relies on a number of critical assumptions, the realism of
which may be disputed.4

3.

Explaining the Z-footnote

Hartwig and Brady (2008) explain the first part of the Z-footnote the proposition that the slope of the Zw-curve equals 1 as follows:
Integrating with respect to N yields: dZ w dZ 1 can be transformed into w. dN dN

dN dN w dN Z w N C

dZ

(1)

Since Z equals the sum of wages and expected profits (see Hartwig and Brady, 2008), the first part of the footnote would make sense if the constant of integration C denoted expected profits. However, normally profits are not constant along Z; they are rising with N.5 What is constant along a linear Z curve is the profit share and therefore, of course, also the wage share because wage income is also a linear function of N in the Z/N-diagram (see Chick 1983, p. 96, for instance). This suggests that Ambrosis conjecture that the footnote has something to do with income shares is indeed correct. This can be shown much more straightforwardly than Ambrosi has done, however. The major flaw in Ambrosis exposition is that he denies that Keynes, in the first part of the footnote, actually states that the slope of Z in wage units (Zw) is equal to 1. But Keynes defines Zw = (N) and concludes: (N) = 1. According to the generally accepted mathematical notation, this means that the derivative of Zw with respect to N, i.e. the slope of the Zw-function in a Zw/N-diagramm, equals 1. Ambrosi disputes the indisputable when he writes: Let us stress once more that the unitary value of the slope is not a statement about the

and Richard Stone who began developing the System of National Accounts during World War II (see Abrahamsen and Hartwig, 2007). One might conjecture that Keyness warnings against measures of aggregate output in the General Theory would have been less pronounced if he had already known the achievements of modern National Accounts.
4

Chick (1983, p. 65f.) mentions the following three assumptions: (i) atomistic firms; (ii) that labour is the only

variable factor of production ; and (iii) that the composition of output and demand does not change with the overall volume of output.
5

There is a special case in which profits are indeed constant along Z, and I will argue below that the first part of

the footnote targets exactly this special case.

shape of the Zw-function itself (Ambrosi, 2011, p. 630). His compelling argument that the last sentence of the footnote states something about the wage share must therefore be derived differently than he has done. It must be derived in a way that respects the wording of the first part of the footnote. Another of Ambrosis insights may be instrumental in doing so, namely the notion that the first part of the footnote and the last sentence deal with different curves. He writes that the footnote refers to characteristics of every individual point which lies on this [aggregate supply] curve but which might at the same time be seen as a point of a different curve namely of a 45-curve which cuts the Zw-curve in every single point of the Zw-curve (Ambrosi, 2011, p. 621). To understand this somewhat obscure sentence, it is probably necessary to have a look at Ambrosis Figure 1, which has a quadrant with a production function of course only for the individual firm because Ambrosi denies the admissibility of the aggregate production function a Zw-curve (also with subscript r indicating the microeconomic level of analysis) and a 45-line which mirrors points on the production function onto the Zwr-curve. Now Ambrosi believes that this 45-line is the function meant by Keynes in the first part of the footnote, and he speaks of Keynes 45-line (although Keynes in the footnote never mentions a 45-line; he just states that the slope the Zw-function without r subscript, one might add equals 1). Although Ambrosis exposition is hence inadequate, his insight that the two parts of the Zfootnote deal with different curves is valuable. It has to be pointed out, however, that this insight has already been published in early 2009 by Claudia Heller (Heller, 2009), who, on her part, builds on Dos Santos Ferreira and Michel (1991). Ambrosi does not mention these contributions. Since Hellers account of the issue is much clearer and more in line with the wording of the footnote than Ambrosis, it will therefore be briefly restated.6 Heller points out that Keynes in the footnote distinguishes between an aggregate supply function Zw = (N) and an aggregate supply curve. This distinction is so subtle that most interpreters including myself have overlooked it so far, believing that the two expressions signify the same thing. Heller, however, draws attention to p. 44 of the General Theory, where Keynes already uses the expression supply curve, which ignoring r subscripts and user cost he defines as:

Z (N ) O (N )

(2)

Dos Santos Ferreira and Michels (1991) focus is not the footnote so I will go along with Heller. With regards

to content, the former reference would be equally adequate.

So the aggregate supply curve is not the same thing as the aggregate supply function (Z). Rearranging (2) yields:

p O Z

(3)

This means that, in terms of chapter 3 of the General Theory, the aggregate supply curve consists of conceivable points of effective demand, i.e. of points of intersection between the aggregate supply function Z and the aggregate demand function D (= p . O), for different conceivable demand price levels p.7 Dos Santos Ferreira and Michel (1991) and Heller (2009) denote this (Z = D)-curve in other words, the aggregate supply curve as Z*. Both Dos Santos Ferreira and Michel (1991) and Heller (2009) acknowledge a resemblance of this curve to the employment function of chapter 20 of the General Theory, which Keynes also defines as a collection of points of effective demand. Now, as Keynes makes clear in chapter 3 of the General Theory, each point of effective demand entails profit maximisation. This implies, as he reiterates in the Z-footnote, that the marginal product must be equal to the marginal factor cost along Z*. This means that the price component implicit in Z* which we can call P* must be equal to the wage rate divided by the marginal product of labour. We therefore know that the Z* and the Zw* curves, respectively, are given by:

Z*

W (N ) * ( N ) Z w '( N ) '( N )

(4)

What we really want to know is the slope of Zw*. Remembering that Z* equals P* . O, we know that Zw* equals (P*/W) . O or Pw* . O. Then we can write (see Arthmar and Brady, 2009, Appendix 3):
* * * d ( Pw O) d ( Pw O) d ( Pw O) 1 '( N ) '( N ) dN dN dO / dN dO

(5)

* Knowing that '( N ) W / P* 1/ Pw because points on Z* are profit-maximizing, we can

rewrite (5) as:

* * d ( Pw O) d ( Pw O) 1 '( N ) * dO dO Pw

(6)

Pw* . O, the effective demand in wage-units, Keynes calls Dw in chapter 20 of the General Theory. So we can rewrite (6) as:
* d ( Pw O) 1 dDw O * dO Pw dO Dw

(7)

For the distinction between the supply price level and the demand price level, see Hartwig (2007), for instance.

On the right-hand side of equation (7) we have the inverse of what Keynes (1936, p. 282f.) defines as the elasticity of output or production. This is therefore the most general expression for the slope of the aggregate supply curve: it equals the inverse of the output elasticity. Under conventional (neo)classical supply-side assumptions, which Keynes accepted in the General Theory I agree with Ambrosi on this point the inverse of the output elasticity equals the inverse of the wage share. So Ambrosis conjecture on the last sentence of the footnote is right, but the result can be derived in a straightforward way, as equations (2) to (7) show. To sum up, there are two errors in the last sentence of the footnote: (i) instead of aggregate supply function it should read aggregate supply curve, and (ii), with respect to the slope of the curve, it should read reciprocal of the money wage share or reciprocal of the output elasticity instead of reciprocal of the money wage. Finally, we can at least partially reconcile the last sentence of the footnote with its first part. (I)f the elasticity of output is unity, writes Keynes (1936, p. 283), no part of the increased effective demand is expected to accrue as profit. From solving equation (1) above backwards, however, we know that in this special case of zero marginal profits, the slope of the aggregate supply function in wage-units (Zw) equals 1. So the statement of the first part of the footnote is correct, but only for the special case of an output elasticity and hence a wage share of 1.

3. Conclusion

Ambrosi (2011) has joined the recent debate over Keyness Z-footnote on pp. 55-56 of the General Theory. He adds two important insights: (i) that the word share has to be inserted at the very end of the footnote in order to make sense of Keyness statement about the slope of the aggregate supply curve, and (ii) that the first part of the footnote and its last sentence are concerned with different functions or curves. However, the originality of (ii) is questionable; and Ambrosis treatment of the first part of the footnote is in contradiction with Keyness explicit statements about the aggregate supply function in wage-units. (i), on its part, can be derived more rigorously than Ambrosi did. To point this out was my first aim with this paper. My second aim was to fight a tendency that otherwise threatens to take root, namely the tendency to use the notion of heterogeneous output in order to deprive macroeconomics of some of the most important and useful concepts it has developed so far. Real GDP is an example for such a concept that scholars like Ambrosi deem inadmissible.

Bibliography

Abrahamsen, Y. and Hartwig, J. 2007. National Accounts, in: Phillip A. OHara (Ed.), International Encyclopedia of Public Policy, Volume 2: Economic Policy. GPERU: Perth, pp. 443-460 Ambrosi, G. M. 2011. Keynes abominable Z-footnote, Cambridge Journal of Economics, vol. 35, no. 3, 619633. Arthmar, R. and Brady, M. E. 2009. Patinkin, Keynes and the Z curve, History of Economic Ideas, vol. 17, no. 3, 127146. Chick, V. 1983. Macroeconomics after Keynes: a Reconsideration of the General Theory, Oxford, Philip Allan. Dos Santos Ferreira, R. and Michel, P. 1991. Keynes aggregate supply function and the principle of effective demand, Recherches Economiques de Louvain, vol. 57, no. 2, 159187. Froyen, R. T. 1976. The aggregative structure of Keyness General Theory, Quarterly Journal of Economics, vol. 90, no. 3, 369387. Hartwig, J. 2007. Keynes versus the Post Keynesians on the principle of effective demand, European Journal of the History of Economic Thought, vol. 14, no. 4, 725739 Hartwig, J. and Brady, M. E. 2008. Comment: Hayes on Z, Cambridge Journal of Economics, vol. 32, no. 5, 815819 Hayes, M. 2007. Keyness Z function, heterogeneous output and marginal productivity, Cambridge Journal of Economics, vol. 31, no. 5, 741753 Hayes, M. 2008. Keyness Z function: a reply to Hartwig and Brady, Cambridge Journal of Economics, vol. 32, no. 5, 811814 Heller, C. 2009. Keyness slip of the pen: aggregate supply curve vs. employment function, MPRA Paper No. 12837 Keynes, J. M. 1936. The General Theory of Employment, Interest, and Money, London, Macmillan.